How are you going to handle sword on sword collisions with regards to feedback?
We've been thinking about this for years. It's not in the videos because to try to explain it here would get us hopelessly deep into the weeds. We think we have an approach that will work. It’s hard to explain in detail without a very lengthy brain dump. It's not just One Big Awesome Solution. It's a number of separate techniques working together. Some of these are familiar (visual, auditory, and haptic feedback) and others center on some innovative UI schemes. If you allow the controller’s position to get out of sync with what is shown on the screen, you get some feedback to that effect and you get UI cues on how to get back into sync.
In general, if you drill down deep enough on the actual sword techniques, the tree of possible outcomes gets pruned way down. It turns out that you rarely have to solve the fully general problem of one sword stopping another sword traveling at top speed at an arbitrary location in space. Which is a hard problem!
If you are "swinging for the fences" with a sword attack---which is to say, if you are assuming a long follow-through---then you're probably doing it wrong. You don't have to cut the other guy in half. You just have to hit him. In most of these arts, you're trained to pull the attack and stop with the sword between you and the adversary. If the attack succeeds, you're done. If it fails, you have stopped with your blade in a tactically sound defensive position instead of swinging all the way through and taking your sword completely out of the action.